Python ML & Computer Vision Engineer
Production Computer Vision · Edge AI Acceleration · Cloud MLOps · LLM & RAG Systems
Python ML Engineer with ~4 years of experience bridging applied Computer Vision with production deployments. Specialized in optimizing edge inference (TensorRT), scaling cloud MLOps platforms, and delivering real-time AI solutions for industrial environments. Actively exploring and applying modern LLM/RAG orchestration to practical projects.
| Category | Technologies & Tools | Scope |
|---|---|---|
| ML & Computer Vision | PyTorch, TensorRT, CUDA, OpenMMLab, YOLO, SAM/FastSAM, OpenCV, GStreamer | Production |
| Backend & DevOps | Python, FastAPI, Docker, Drone CI, GitHub Actions, Integration Tests, AWS/GCP | Production |
| Simulation | Gazebo, Webots | Production / Project |
| LLM & RAG Systems | LangChain, LangGraph, LlamaIndex, MCP, FAISS, ChromaDB, Ollama | Applied Projects |
- Cloud Pipelines & CV Modalities: Architected 1-click GPU training workflows and custom-forked Label Studio for 4 CV modalities, slashing model retraining from 8 hours to 30 minutes.
- AI-Assisted Labeling & Team Sync: Coordinated an 8-member cross-functional squad to establish release cycles; integrated SAM/FastSAM, reducing manual annotation overhead by 30%.
- Real-Time Edge Deployment: Optimized deep learning models using TensorRT (FP16/INT8) for NVIDIA Jetson/Orin, meeting strict real-time inference requirements during on-premise deployments for tier-1 industrial clients.
- Synthetic Data: Built 3D warehouse simulations (Gazebo, Webots) to generate synthetic datasets, automating keypoint model retraining loops without manual data collection.
- 🛡️ Kivy-App: Secure desktop application for real-time YOLO object detection. Features AES-256 local database encryption, headless UI testing (Xvfb), and pytest CI/CD pipelines.
- 🧠 Explainable-AI-MRNet: Explainable deep learning framework (CAM, SHAP) evaluated on knee MRI scans. Source code for JAISCR (2024) publication.
- 📊 Browser Utilities: Zero-dependency standalone JS tools for analytics (ReadingsTracker) and interactive CSV parsing (CardMatch).
- 🔬 A Novel Explainable AI Model for Medical Data Analysis (JAISCR, 2024) — Hybrid ensemble architecture for high-dimensional imaging. [Publisher] [Offline PDF] [Code]
- 🔬 Architecture of Document OCR System (Herald Tech. Sci., 2022) — Multi-language OCR pipeline. [Publisher] [Offline PDF]
- 🎓 M.Sc. AI & B.Sc. CS — Lviv Polytechnic National University (ZAB Statements of Comparability available).
- 🗣️ Languages: English (Professional / B2+), German (B2 in progress), Ukrainian (Native).